RNG Testing Overview - How It Will Work
General Overview The RNG Testing that I am performing is actually a test of the Random Number Generator and Match Maker which are implemented within World of Tanks. In order to settle the contraversy of whether these two systems are rigged to favor lower-skilled players or not, I have decided to undertake a massive thurough analysis of the two. 5 million data points will be collected and professionally analyzed, in order to ensure that the testing will suffice to prove one of the two sides of the debate once and for all. Meathods of Testing The testing will be a multi-step process. The primary step in the testing is comparing player's RNG rolls on pre-set, uniform tanks. This will be done by analyzing how the Win Rate of a shooter and target affect the RNG of the shot, if it has any effect at all. The following is the basic pattern that the testing will follow. (Shooter's Win Rate ---> Target's Win Rate) 40%-42% ---> 40%-42% 40%-42% ---> 43%-45% 40%-42% ---> 46%-48% 40%-42% ---> 49%-51% 40%-42% ---> 52%-54% 40%-42% ---> 55%-59% 40%-42% ---> 60%+ 43%-45% ---> 40%-42% 43%-45% ---> 43%-45% 43%-45% ---> 46%-48% 43%-45% ---> 49%-51% 43%-45% ---> 52%-54% 43%-45% ---> 55%-59% 43%-45% ---> 60%+ 46%-48% ---> 40%-42% 46%-48% ---> 43%-45% 46%-48% ---> 46%-48% 46%-48% ---> 49%-51% 46%-48% ---> 52%-54% 46%-48% ---> 55%-59% 46%-48% ---> 60%+ 49%-51% ---> 40%-42% 49%-51% ---> 43%-45% 49%-51% ---> 46%-48% 49%-51% ---> 49%-51% 49%-51% ---> 52%-54% 49%-51% ---> 55%-59% 49%-51% ---> 60%+ 52%-54% ---> 40%-42% 52%-54% ---> 43%-45% 52%-54% ---> 46%-48% 52%-54% ---> 49%-51% 52%-54% ---> 52%-54% 52%-54% ---> 55%-59% 52%-54% ---> 60%+ 55%-59% ---> 40%-42% 55%-59% ---> 43%-45% 55%-59% ---> 46%-48% 55%-59% ---> 49%-51% 55%-59% ---> 52%-54% 55%-59% ---> 55%-59% 55%-59% ---> 60%+ 60%+ ---> 40%-42% 60%+ ---> 43%-45% 60%+ ---> 46%-48% 60%+ ---> 49%-51% 60%+ ---> 52%-54% 60%+ ---> 55%-59% 60%+ ---> 60%+ For the very basics of the testing, each of the shooters will shoot 100 bullets into a target of each of the above Win Rates. There will be 5 shooters and 5 targets for each Win Rate category involved, so with 5 shooters of 7 different Win Rates, shooting 100 bullets into targets of 7 different Win Rates, that's a total of 24,500 data points that will be collected for this first portion of the testing. But of course the testing goes into much more detail than that. The next part of the testing will involve Gun Elevation and Depression Angles. Using a special mod, we will ensure that we elevate/depress the guns by increments of 2''' degrees each time. This means that the following gun elevation/depression angles will be tested. -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 16 18 20 Each of the shooters will shoot 30 bullets at '''each of these elevation/depression angles. Using the same Win Rate on Win Rate system, we're looking at 30 bullets for each of the 16 gun elevation/depression angles, with 5 shooters from each of the 7 different Win Rates categories, shooting at targets form the 7 different Win Rate categories. In otherwords, 117,600 points of data. The next part of the testing involves tracking module damage. Using another special modification which displays the locations of a tank's modules and crew members, we will be tracking module damage by shooting 50 rounds into each of a tank's hit point zones. This will give us an additional 98,000 data points. Chances of fire will be tested as well, but in a different manner than I had originally stated. I will be testing a theoretical chance of fire of 5%, 10%, 15%, and 20%. Shooters will unload 30 shots into the targets' engines for each of the theoretical chances of fire. Shooters will do the same thing for engines of each of the 7 different nations. 205,800 data points here. Penetration! I'll be testing the damage dealt by shots of various penetration values and I'll be tracking simply what kind of penetration values players get. Tanks with an effective armor within a 10mm range of various points in the T49's possible penetration values will be used, meaning: (96mm-106mm) (107mm-117mm) (118mm-128mm) (129mm-139mm) (140mm-150mm) (151mm-160mm). Shooters will propel 100 shots at each of these armor values. 147,000 shots will be recorded during this part of the testing. Premium Accounts will also be tested separately. I'll group together 3 users from each of the 7 Win Rate categories. Each of these users will shoot 150 rounds into enemy vehicles in Random Battles. I'll average up the damage dealt values from the following two categories of win rate groups: High: (52%-54%) (55%-59%) (60%+) Low: (40%-42%) (43%-45%) (46%-48%) (49%-51%) I will then compare the results and see how the RNG variability was affected, if at all. Only 3,150 shots will be tracked for this; as I don't want to spend too much time on it. Random Battles will also be tested by specially designated Battle Recorders. One from each Win Rate group will do 500 battles in their T49 and send all 500 replays to the Coordinators. They will have 2 months to do so. This will help us get real Random Battle data as well. The number of bits of data that will be recorded here is unknown, but with a total of 3,500 battles that will be done for this part of the testing, we can be sure that it will be immense! I am greatly aware that during those 500 battles, a player's Win Rate would have a great opportunity to change. If anyone has any ideas how to fix this issue, please let me know in the Wikia Forums! And last but not least for RNG, accuracy! 100 shots will be propelled each at the following chances to hit the target: 100%, 80%, 60%, 40%, 20%, 5%. This will result in a total of 147,000 shots fired. Now! For! The! Match Maker! I'll designate 5 players of varying Win Rates to send me 500 Replays of their Match Ups for Tier 3 battles, Tier 6 battles, and Tier 9 battles. This means that each player will be sending me 1,500 replays, for a total of 7,500 Replays. I'll then analyze the Match-Ups in two ways. The first will simply be the tank balancing on each side (I.E. How many Tier 6's, 7's and 8's each side has, etc.). Then I will compare the stats of the players on each team from the Match-Ups, and then go ahead and proceed and analyze the Chance to Wins that XVM gives, and the Chance to Win that I give the Match-Up (I have my own system for determining this that takes into account more factors than just skill). I'll then post results along with the 9.0 RNG Testing Results.